Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 11 aaronmk
import re
5 3238 aaronmk
import time
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 3241 aaronmk
import profiling
13 1889 aaronmk
from Proxy import Proxy
14 1872 aaronmk
import rand
15 2217 aaronmk
import sql_gen
16 862 aaronmk
import strings
17 131 aaronmk
import util
18 11 aaronmk
19 832 aaronmk
##### Exceptions
20
21 2804 aaronmk
def get_cur_query(cur, input_query=None):
22 2168 aaronmk
    raw_query = None
23
    if hasattr(cur, 'query'): raw_query = cur.query
24
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
25 2170 aaronmk
26
    if raw_query != None: return raw_query
27 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
28 14 aaronmk
29 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
30
    '''For params, see get_cur_query()'''
31 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(get_cur_query(*args, **kw_args)))
32 135 aaronmk
33 300 aaronmk
class DbException(exc.ExceptionWithCause):
34 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
35 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
36 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
37
38 2143 aaronmk
class ExceptionWithName(DbException):
39
    def __init__(self, name, cause=None):
40 4491 aaronmk
        DbException.__init__(self, 'for name: '
41
            +strings.as_tt(strings.ustr(name)), cause)
42 2143 aaronmk
        self.name = name
43 360 aaronmk
44 3109 aaronmk
class ExceptionWithValue(DbException):
45
    def __init__(self, value, cause=None):
46 4492 aaronmk
        DbException.__init__(self, 'for value: '
47
            +strings.as_tt(strings.urepr(value)), cause)
48 2240 aaronmk
        self.value = value
49
50 2945 aaronmk
class ExceptionWithNameType(DbException):
51
    def __init__(self, type_, name, cause=None):
52 4491 aaronmk
        DbException.__init__(self, 'for type: '+strings.as_tt(strings.ustr(
53
            type_))+'; name: '+strings.as_tt(name), cause)
54 2945 aaronmk
        self.type = type_
55
        self.name = name
56
57 2306 aaronmk
class ConstraintException(DbException):
58 3345 aaronmk
    def __init__(self, name, cond, cols, cause=None):
59
        msg = 'Violated '+strings.as_tt(name)+' constraint'
60 5447 aaronmk
        if cond != None: msg += ' with condition '+strings.as_tt(cond)
61 3345 aaronmk
        if cols != []: msg += ' on columns: '+strings.as_tt(', '.join(cols))
62
        DbException.__init__(self, msg, cause)
63 2306 aaronmk
        self.name = name
64 3345 aaronmk
        self.cond = cond
65 468 aaronmk
        self.cols = cols
66 11 aaronmk
67 2523 aaronmk
class MissingCastException(DbException):
68 4139 aaronmk
    def __init__(self, type_, col=None, cause=None):
69
        msg = 'Missing cast to type '+strings.as_tt(type_)
70
        if col != None: msg += ' on column: '+strings.as_tt(col)
71
        DbException.__init__(self, msg, cause)
72 2523 aaronmk
        self.type = type_
73
        self.col = col
74
75 5576 aaronmk
class EncodingException(ExceptionWithName): pass
76
77 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
78 13 aaronmk
79 2306 aaronmk
class NullValueException(ConstraintException): pass
80 13 aaronmk
81 3346 aaronmk
class CheckException(ConstraintException): pass
82
83 3109 aaronmk
class InvalidValueException(ExceptionWithValue): pass
84 2239 aaronmk
85 2945 aaronmk
class DuplicateException(ExceptionWithNameType): pass
86 2143 aaronmk
87 3419 aaronmk
class DoesNotExistException(ExceptionWithNameType): pass
88
89 89 aaronmk
class EmptyRowException(DbException): pass
90
91 865 aaronmk
##### Warnings
92
93
class DbWarning(UserWarning): pass
94
95 1930 aaronmk
##### Result retrieval
96
97
def col_names(cur): return (col[0] for col in cur.description)
98
99
def rows(cur): return iter(lambda: cur.fetchone(), None)
100
101
def consume_rows(cur):
102
    '''Used to fetch all rows so result will be cached'''
103
    iters.consume_iter(rows(cur))
104
105
def next_row(cur): return rows(cur).next()
106
107
def row(cur):
108
    row_ = next_row(cur)
109
    consume_rows(cur)
110
    return row_
111
112
def next_value(cur): return next_row(cur)[0]
113
114
def value(cur): return row(cur)[0]
115
116
def values(cur): return iters.func_iter(lambda: next_value(cur))
117
118
def value_or_none(cur):
119
    try: return value(cur)
120
    except StopIteration: return None
121
122 2762 aaronmk
##### Escaping
123 2101 aaronmk
124 2573 aaronmk
def esc_name_by_module(module, name):
125
    if module == 'psycopg2' or module == None: quote = '"'
126 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
127
    else: raise NotImplementedError("Can't escape name for "+module+' database')
128 2500 aaronmk
    return sql_gen.esc_name(name, quote)
129 2101 aaronmk
130
def esc_name_by_engine(engine, name, **kw_args):
131
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
132
133
def esc_name(db, name, **kw_args):
134
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
135
136
def qual_name(db, schema, table):
137
    def esc_name_(name): return esc_name(db, name)
138
    table = esc_name_(table)
139
    if schema != None: return esc_name_(schema)+'.'+table
140
    else: return table
141
142 1869 aaronmk
##### Database connections
143 1849 aaronmk
144 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
145 1926 aaronmk
146 1869 aaronmk
db_engines = {
147
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
148
    'PostgreSQL': ('psycopg2', {}),
149
}
150
151
DatabaseErrors_set = set([DbException])
152
DatabaseErrors = tuple(DatabaseErrors_set)
153
154
def _add_module(module):
155
    DatabaseErrors_set.add(module.DatabaseError)
156
    global DatabaseErrors
157
    DatabaseErrors = tuple(DatabaseErrors_set)
158
159
def db_config_str(db_config):
160
    return db_config['engine']+' database '+db_config['database']
161
162 2448 aaronmk
log_debug_none = lambda msg, level=2: None
163 1901 aaronmk
164 1849 aaronmk
class DbConn:
165 2923 aaronmk
    def __init__(self, db_config, autocommit=True, caching=True,
166 3183 aaronmk
        log_debug=log_debug_none, debug_temp=False, src=None):
167 2915 aaronmk
        '''
168
        @param debug_temp Whether temporary objects should instead be permanent.
169
            This assists in debugging the internal objects used by the program.
170 3183 aaronmk
        @param src In autocommit mode, will be included in a comment in every
171
            query, to help identify the data source in pg_stat_activity.
172 2915 aaronmk
        '''
173 1869 aaronmk
        self.db_config = db_config
174 2190 aaronmk
        self.autocommit = autocommit
175
        self.caching = caching
176 1901 aaronmk
        self.log_debug = log_debug
177 2193 aaronmk
        self.debug = log_debug != log_debug_none
178 2915 aaronmk
        self.debug_temp = debug_temp
179 3183 aaronmk
        self.src = src
180 3074 aaronmk
        self.autoanalyze = False
181 3269 aaronmk
        self.autoexplain = False
182
        self.profile_row_ct = None
183 1869 aaronmk
184 3124 aaronmk
        self._savepoint = 0
185 3120 aaronmk
        self._reset()
186 1869 aaronmk
187
    def __getattr__(self, name):
188
        if name == '__dict__': raise Exception('getting __dict__')
189
        if name == 'db': return self._db()
190
        else: raise AttributeError()
191
192
    def __getstate__(self):
193
        state = copy.copy(self.__dict__) # shallow copy
194 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
195 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
196
        return state
197
198 3118 aaronmk
    def clear_cache(self): self.query_results = {}
199
200 3120 aaronmk
    def _reset(self):
201 3118 aaronmk
        self.clear_cache()
202 3124 aaronmk
        assert self._savepoint == 0
203 3118 aaronmk
        self._notices_seen = set()
204
        self.__db = None
205
206 2165 aaronmk
    def connected(self): return self.__db != None
207
208 3116 aaronmk
    def close(self):
209 3119 aaronmk
        if not self.connected(): return
210
211 3135 aaronmk
        # Record that the automatic transaction is now closed
212 3136 aaronmk
        self._savepoint -= 1
213 3135 aaronmk
214 3119 aaronmk
        self.db.close()
215 3120 aaronmk
        self._reset()
216 3116 aaronmk
217 3125 aaronmk
    def reconnect(self):
218
        # Do not do this in test mode as it would roll back everything
219
        if self.autocommit: self.close()
220
        # Connection will be reopened automatically on first query
221
222 1869 aaronmk
    def _db(self):
223
        if self.__db == None:
224
            # Process db_config
225
            db_config = self.db_config.copy() # don't modify input!
226 2097 aaronmk
            schemas = db_config.pop('schemas', None)
227 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
228
            module = __import__(module_name)
229
            _add_module(module)
230
            for orig, new in mappings.iteritems():
231
                try: util.rename_key(db_config, orig, new)
232
                except KeyError: pass
233
234
            # Connect
235
            self.__db = module.connect(**db_config)
236
237 3161 aaronmk
            # Record that a transaction is already open
238
            self._savepoint += 1
239
240 1869 aaronmk
            # Configure connection
241 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
242
                import psycopg2.extensions
243
                self.db.set_isolation_level(
244
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
245 2101 aaronmk
            if schemas != None:
246 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
247
                search_path.append(value(run_query(self, 'SHOW search_path',
248
                    log_level=4)))
249
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
250
                    log_level=3)
251 1869 aaronmk
252
        return self.__db
253 1889 aaronmk
254 1891 aaronmk
    class DbCursor(Proxy):
255 1927 aaronmk
        def __init__(self, outer):
256 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
257 2191 aaronmk
            self.outer = outer
258 1927 aaronmk
            self.query_results = outer.query_results
259 1894 aaronmk
            self.query_lookup = None
260 1891 aaronmk
            self.result = []
261 1889 aaronmk
262 2802 aaronmk
        def execute(self, query):
263 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
264 2797 aaronmk
            self.query_lookup = query
265 2148 aaronmk
            try:
266 3162 aaronmk
                try: cur = self.inner.execute(query)
267 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
268 1904 aaronmk
            except Exception, e:
269
                self.result = e # cache the exception as the result
270
                self._cache_result()
271
                raise
272 3004 aaronmk
273
            # Always cache certain queries
274 3183 aaronmk
            query = sql_gen.lstrip(query)
275 3004 aaronmk
            if query.startswith('CREATE') or query.startswith('ALTER'):
276 3007 aaronmk
                # structural changes
277 3040 aaronmk
                # Rest of query must be unique in the face of name collisions,
278
                # so don't cache ADD COLUMN unless it has distinguishing comment
279
                if query.find('ADD COLUMN') < 0 or query.endswith('*/'):
280 3007 aaronmk
                    self._cache_result()
281 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
282 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
283 3004 aaronmk
284 2762 aaronmk
            return cur
285 1894 aaronmk
286 1891 aaronmk
        def fetchone(self):
287
            row = self.inner.fetchone()
288 1899 aaronmk
            if row != None: self.result.append(row)
289
            # otherwise, fetched all rows
290 1904 aaronmk
            else: self._cache_result()
291
            return row
292
293
        def _cache_result(self):
294 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
295
            # inserts are not idempotent. Other non-SELECT queries don't have
296
            # their result set read, so only exceptions will be cached (an
297
            # invalid query will always be invalid).
298 1930 aaronmk
            if self.query_results != None and (not self._is_insert
299 1906 aaronmk
                or isinstance(self.result, Exception)):
300
301 1894 aaronmk
                assert self.query_lookup != None
302 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
303
                    util.dict_subset(dicts.AttrsDictView(self),
304
                    ['query', 'result', 'rowcount', 'description']))
305 1906 aaronmk
306 1916 aaronmk
        class CacheCursor:
307
            def __init__(self, cached_result): self.__dict__ = cached_result
308
309 1927 aaronmk
            def execute(self, *args, **kw_args):
310 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
311
                # otherwise, result is a rows list
312
                self.iter = iter(self.result)
313
314
            def fetchone(self):
315
                try: return self.iter.next()
316
                except StopIteration: return None
317 1891 aaronmk
318 2212 aaronmk
    def esc_value(self, value):
319 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
320
        except NotImplementedError, e:
321
            module = util.root_module(self.db)
322
            if module == 'MySQLdb':
323
                import _mysql
324
                str_ = _mysql.escape_string(value)
325
            else: raise e
326 2374 aaronmk
        return strings.to_unicode(str_)
327 2212 aaronmk
328 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
329
330 2814 aaronmk
    def std_code(self, str_):
331
        '''Standardizes SQL code.
332
        * Ensures that string literals are prefixed by `E`
333
        '''
334
        if str_.startswith("'"): str_ = 'E'+str_
335
        return str_
336
337 2665 aaronmk
    def can_mogrify(self):
338 2663 aaronmk
        module = util.root_module(self.db)
339 2665 aaronmk
        return module == 'psycopg2'
340 2663 aaronmk
341 2665 aaronmk
    def mogrify(self, query, params=None):
342
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
343
        else: raise NotImplementedError("Can't mogrify query")
344
345 5579 aaronmk
    def set_encoding(self, encoding):
346
        encoding_str = sql_gen.Literal(encoding)
347
        run_query(self, 'SET NAMES '+encoding_str.to_str(self))
348
349 2671 aaronmk
    def print_notices(self):
350 2725 aaronmk
        if hasattr(self.db, 'notices'):
351
            for msg in self.db.notices:
352
                if msg not in self._notices_seen:
353
                    self._notices_seen.add(msg)
354
                    self.log_debug(msg, level=2)
355 2671 aaronmk
356 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
357 2464 aaronmk
        debug_msg_ref=None):
358 2445 aaronmk
        '''
359 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
360
            throws one of these exceptions.
361 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
362
            this instead of being output. This allows you to filter log messages
363
            depending on the result of the query.
364 2445 aaronmk
        '''
365 2167 aaronmk
        assert query != None
366
367 3183 aaronmk
        if self.autocommit and self.src != None:
368 3206 aaronmk
            query = sql_gen.esc_comment(self.src)+'\t'+query
369 3183 aaronmk
370 2047 aaronmk
        if not self.caching: cacheable = False
371 1903 aaronmk
        used_cache = False
372 2664 aaronmk
373 3242 aaronmk
        if self.debug:
374
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
375 1903 aaronmk
        try:
376 1927 aaronmk
            # Get cursor
377
            if cacheable:
378 3238 aaronmk
                try: cur = self.query_results[query]
379 1927 aaronmk
                except KeyError: cur = self.DbCursor(self)
380 3238 aaronmk
                else: used_cache = True
381 1927 aaronmk
            else: cur = self.db.cursor()
382
383
            # Run query
384 3238 aaronmk
            try: cur.execute(query)
385 3162 aaronmk
            except Exception, e:
386
                _add_cursor_info(e, self, query)
387
                raise
388 3238 aaronmk
            else: self.do_autocommit()
389 1903 aaronmk
        finally:
390 3242 aaronmk
            if self.debug:
391 3244 aaronmk
                profiler.stop(self.profile_row_ct)
392 3242 aaronmk
393
                ## Log or return query
394
395 4491 aaronmk
                query = strings.ustr(get_cur_query(cur, query))
396 3281 aaronmk
                # Put the src comment on a separate line in the log file
397
                query = query.replace('\t', '\n', 1)
398 3239 aaronmk
399 3240 aaronmk
                msg = 'DB query: '
400 3239 aaronmk
401 3240 aaronmk
                if used_cache: msg += 'cache hit'
402
                elif cacheable: msg += 'cache miss'
403
                else: msg += 'non-cacheable'
404 3239 aaronmk
405 3241 aaronmk
                msg += ':\n'+profiler.msg()+'\n'+strings.as_code(query, 'SQL')
406 3240 aaronmk
407 3237 aaronmk
                if debug_msg_ref != None: debug_msg_ref[0] = msg
408
                else: self.log_debug(msg, log_level)
409 3245 aaronmk
410
                self.print_notices()
411 1903 aaronmk
412
        return cur
413 1914 aaronmk
414 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
415 2139 aaronmk
416 2907 aaronmk
    def with_autocommit(self, func):
417 2801 aaronmk
        import psycopg2.extensions
418
419
        prev_isolation_level = self.db.isolation_level
420 2907 aaronmk
        self.db.set_isolation_level(
421
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
422 2683 aaronmk
        try: return func()
423 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
424 2683 aaronmk
425 2139 aaronmk
    def with_savepoint(self, func):
426 3137 aaronmk
        top = self._savepoint == 0
427 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
428 3137 aaronmk
429 3272 aaronmk
        if self.debug:
430 3273 aaronmk
            self.log_debug('Begin transaction', level=4)
431 3272 aaronmk
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
432
433 3160 aaronmk
        # Must happen before running queries so they don't get autocommitted
434
        self._savepoint += 1
435
436 3137 aaronmk
        if top: query = 'START TRANSACTION ISOLATION LEVEL READ COMMITTED'
437
        else: query = 'SAVEPOINT '+savepoint
438
        self.run_query(query, log_level=4)
439
        try:
440
            return func()
441
            if top: self.run_query('COMMIT', log_level=4)
442 2139 aaronmk
        except:
443 3137 aaronmk
            if top: query = 'ROLLBACK'
444
            else: query = 'ROLLBACK TO SAVEPOINT '+savepoint
445
            self.run_query(query, log_level=4)
446
447 2139 aaronmk
            raise
448 2930 aaronmk
        finally:
449
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
450
            # "The savepoint remains valid and can be rolled back to again"
451
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
452 3137 aaronmk
            if not top:
453
                self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
454 2930 aaronmk
455
            self._savepoint -= 1
456
            assert self._savepoint >= 0
457
458 3272 aaronmk
            if self.debug:
459
                profiler.stop(self.profile_row_ct)
460 3273 aaronmk
                self.log_debug('End transaction\n'+profiler.msg(), level=4)
461 3272 aaronmk
462 2930 aaronmk
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
463 2191 aaronmk
464
    def do_autocommit(self):
465
        '''Autocommits if outside savepoint'''
466 3135 aaronmk
        assert self._savepoint >= 1
467
        if self.autocommit and self._savepoint == 1:
468 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
469 2191 aaronmk
            self.db.commit()
470 2643 aaronmk
471 3155 aaronmk
    def col_info(self, col, cacheable=True):
472 2643 aaronmk
        table = sql_gen.Table('columns', 'information_schema')
473 4936 aaronmk
        cols = [sql_gen.Col('data_type'), sql_gen.Col('udt_name'),
474
            'column_default', sql_gen.Cast('boolean',
475
            sql_gen.Col('is_nullable'))]
476 2643 aaronmk
477 3750 aaronmk
        conds = [('table_name', col.table.name),
478
            ('column_name', strings.ustr(col.name))]
479 2643 aaronmk
        schema = col.table.schema
480
        if schema != None: conds.append(('table_schema', schema))
481
482 3638 aaronmk
        cur = select(self, table, cols, conds, order_by='table_schema', limit=1,
483
            cacheable=cacheable, log_level=4) # TODO: order by search_path order
484 4936 aaronmk
        try: type_, extra_type, default, nullable = row(cur)
485 4114 aaronmk
        except StopIteration: raise sql_gen.NoUnderlyingTableException(col)
486 2819 aaronmk
        default = sql_gen.as_Code(default, self)
487 4936 aaronmk
        if type_ == 'USER-DEFINED': type_ = extra_type
488 4939 aaronmk
        elif type_ == 'ARRAY':
489
            type_ = sql_gen.ArrayType(strings.remove_prefix('_', extra_type,
490
                require=True))
491 2819 aaronmk
492
        return sql_gen.TypedCol(col.name, type_, default, nullable)
493 2917 aaronmk
494
    def TempFunction(self, name):
495
        if self.debug_temp: schema = None
496
        else: schema = 'pg_temp'
497
        return sql_gen.Function(name, schema)
498 1849 aaronmk
499 1869 aaronmk
connect = DbConn
500
501 832 aaronmk
##### Recoverable querying
502 15 aaronmk
503 5577 aaronmk
def parse_exception(db, e, recover=False):
504
    msg = strings.ustr(e.args[0])
505
    msg = re.sub(r'^(?:PL/Python: )?ValueError: ', r'', msg)
506
507
    match = re.match(r'^invalid byte sequence for encoding "(.+?)":', msg)
508
    if match:
509
        encoding, = match.groups()
510
        raise EncodingException(encoding, e)
511
512 5763 aaronmk
    def make_DuplicateKeyException(constraint, e):
513 5577 aaronmk
        cols = []
514
        cond = None
515
        if recover: # need auto-rollback to run index_cols()
516
            try:
517
                cols = index_cols(db, constraint)
518
                cond = index_cond(db, constraint)
519
            except NotImplementedError: pass
520 5763 aaronmk
        return DuplicateKeyException(constraint, cond, cols, e)
521 5577 aaronmk
522 5763 aaronmk
    match = re.match(r'^duplicate key value violates unique constraint "(.+?)"',
523
        msg)
524
    if match:
525
        constraint, = match.groups()
526
        raise make_DuplicateKeyException(constraint, e)
527
528 5764 aaronmk
    match = re.match(r'^could not create unique index "(.+?)"\n'
529
        r'DETAIL:  Key .+? is duplicated', msg)
530
    if match:
531
        constraint, = match.groups()
532 5823 aaronmk
        raise DuplicateKeyException(constraint, None, [], e)
533 5764 aaronmk
534 5577 aaronmk
    match = re.match(r'^null value in column "(.+?)" violates not-null'
535
        r' constraint', msg)
536
    if match:
537
        col, = match.groups()
538
        raise NullValueException('NOT NULL', None, [col], e)
539
540
    match = re.match(r'^new row for relation "(.+?)" violates check '
541
        r'constraint "(.+?)"', msg)
542
    if match:
543
        table, constraint = match.groups()
544
        constraint = sql_gen.Col(constraint, table)
545
        cond = None
546
        if recover: # need auto-rollback to run constraint_cond()
547
            try: cond = constraint_cond(db, constraint)
548
            except NotImplementedError: pass
549
        raise CheckException(constraint.to_str(db), cond, [], e)
550
551 5723 aaronmk
    match = re.match(r'^(?:invalid input (?:syntax|value)\b[^:]*'
552 5717 aaronmk
        r'|.+? out of range)(?:: "(.+?)")?', msg)
553 5577 aaronmk
    if match:
554
        value, = match.groups()
555 5717 aaronmk
        value = util.do_ignore_none(strings.to_unicode, value)
556
        raise InvalidValueException(value, e)
557 5577 aaronmk
558
    match = re.match(r'^column "(.+?)" is of type (.+?) but expression '
559
        r'is of type', msg)
560
    if match:
561
        col, type_ = match.groups()
562
        raise MissingCastException(type_, col, e)
563
564
    match = re.match(r'^could not determine polymorphic type because '
565
        r'input has type "unknown"', msg)
566
    if match: raise MissingCastException('text', None, e)
567
568 5724 aaronmk
    match = re.match(r'^.+? types (.+?) and (.+?) cannot be matched', msg)
569
    if match:
570
        type0, type1 = match.groups()
571
        raise MissingCastException(type0, None, e)
572 5577 aaronmk
573 5707 aaronmk
    typed_name_re = r'^(\S+) "?(.+?)"?(?: of relation ".+?")?'
574 5577 aaronmk
575
    match = re.match(typed_name_re+r'.*? already exists', msg)
576
    if match:
577
        type_, name = match.groups()
578
        raise DuplicateException(type_, name, e)
579
580
    match = re.match(r'more than one (\S+) named ""(.+?)""', msg)
581
    if match:
582
        type_, name = match.groups()
583
        raise DuplicateException(type_, name, e)
584
585
    match = re.match(typed_name_re+r' does not exist', msg)
586
    if match:
587
        type_, name = match.groups()
588 5683 aaronmk
        if type_ == 'function':
589 5709 aaronmk
            match = re.match(r'^(.+?)\(.*\)$', name)
590 5710 aaronmk
            if match: # includes params, so is call rather than cast to regproc
591 5708 aaronmk
                function_name, = match.groups()
592 5711 aaronmk
                func = sql_gen.Function(function_name)
593 5712 aaronmk
                if function_exists(db, func) and msg.find('CAST') < 0:
594 5683 aaronmk
                    # not found only because of a missing cast
595 5715 aaronmk
                    type_ = function_param0_type(db, func)
596 5792 aaronmk
                    if type_ == 'anyelement': type_ = 'text'
597 5715 aaronmk
                    raise MissingCastException(type_, None, e)
598 5577 aaronmk
        raise DoesNotExistException(type_, name, e)
599
600
    raise # no specific exception raised
601
602 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
603 11 aaronmk
604 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
605
    log_ignore_excs=None, **kw_args):
606 2794 aaronmk
    '''For params, see DbConn.run_query()'''
607 830 aaronmk
    if recover == None: recover = False
608 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
609
    log_ignore_excs = tuple(log_ignore_excs)
610 3236 aaronmk
    debug_msg_ref = [None]
611 830 aaronmk
612 3267 aaronmk
    query = with_explain_comment(db, query)
613 3258 aaronmk
614 2148 aaronmk
    try:
615 2464 aaronmk
        try:
616 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
617 2793 aaronmk
                debug_msg_ref, **kw_args)
618 2796 aaronmk
            if recover and not db.is_cached(query):
619 2464 aaronmk
                return with_savepoint(db, run)
620
            else: return run() # don't need savepoint if cached
621 5890 aaronmk
        except Exception, e:
622
            # Give failed EXPLAIN approximately the log_level of its query
623
            if query.startswith('EXPLAIN'): log_level -= 1
624
625
            parse_exception(db, e, recover)
626 2464 aaronmk
    except log_ignore_excs:
627
        log_level += 2
628
        raise
629
    finally:
630 3236 aaronmk
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
631 830 aaronmk
632 832 aaronmk
##### Basic queries
633
634 3256 aaronmk
def is_explainable(query):
635
    # See <http://www.postgresql.org/docs/8.3/static/sql-explain.html#AEN57749>
636 3257 aaronmk
    return re.match(r'^(?:SELECT|INSERT|UPDATE|DELETE|VALUES|EXECUTE|DECLARE)\b'
637
        , query)
638 3256 aaronmk
639 3263 aaronmk
def explain(db, query, **kw_args):
640
    '''
641
    For params, see run_query().
642
    '''
643 3267 aaronmk
    kw_args.setdefault('log_level', 4)
644 3263 aaronmk
645 3750 aaronmk
    return strings.ustr(strings.join_lines(values(run_query(db,
646
        'EXPLAIN '+query, recover=True, cacheable=True, **kw_args))))
647 3256 aaronmk
        # not a higher log_level because it's useful to see what query is being
648
        # run before it's executed, which EXPLAIN effectively provides
649
650 3265 aaronmk
def has_comment(query): return query.endswith('*/')
651
652
def with_explain_comment(db, query, **kw_args):
653 3269 aaronmk
    if db.autoexplain and not has_comment(query) and is_explainable(query):
654 3265 aaronmk
        query += '\n'+sql_gen.esc_comment(' EXPLAIN:\n'
655
            +explain(db, query, **kw_args))
656
    return query
657
658 2153 aaronmk
def next_version(name):
659 2163 aaronmk
    version = 1 # first existing name was version 0
660 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
661 2153 aaronmk
    if match:
662 2586 aaronmk
        name, version = match.groups()
663
        version = int(version)+1
664 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
665 2153 aaronmk
666 2899 aaronmk
def lock_table(db, table, mode):
667
    table = sql_gen.as_Table(table)
668
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
669
670 3303 aaronmk
def run_query_into(db, query, into=None, add_pkey_=False, **kw_args):
671 2085 aaronmk
    '''Outputs a query to a temp table.
672
    For params, see run_query().
673
    '''
674 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
675 2790 aaronmk
676
    assert isinstance(into, sql_gen.Table)
677
678 2992 aaronmk
    into.is_temp = True
679 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
680
    into.schema = None
681 2992 aaronmk
682 2790 aaronmk
    kw_args['recover'] = True
683 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
684 2790 aaronmk
685 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
686 2790 aaronmk
687
    # Create table
688
    while True:
689
        create_query = 'CREATE'
690
        if temp: create_query += ' TEMP'
691
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
692 2385 aaronmk
693 2790 aaronmk
        try:
694
            cur = run_query(db, create_query, **kw_args)
695
                # CREATE TABLE AS sets rowcount to # rows in query
696
            break
697 2945 aaronmk
        except DuplicateException, e:
698 2790 aaronmk
            into.name = next_version(into.name)
699
            # try again with next version of name
700
701 3303 aaronmk
    if add_pkey_: add_pkey(db, into)
702 3075 aaronmk
703
    # According to the PostgreSQL doc, "The autovacuum daemon cannot access and
704
    # therefore cannot vacuum or analyze temporary tables. [...] if a temporary
705
    # table is going to be used in complex queries, it is wise to run ANALYZE on
706
    # the temporary table after it is populated."
707
    # (http://www.postgresql.org/docs/9.1/static/sql-createtable.html)
708
    # If into is not a temp table, ANALYZE is useful but not required.
709 3073 aaronmk
    analyze(db, into)
710 2790 aaronmk
711
    return cur
712 2085 aaronmk
713 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
714
715 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
716
717 3420 aaronmk
def mk_select(db, tables=None, fields=None, conds=None, distinct_on=[],
718 3494 aaronmk
    limit=None, start=None, order_by=order_by_pkey, default_table=None,
719
    explain=True):
720 1981 aaronmk
    '''
721 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
722 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
723 1981 aaronmk
    @param fields Use None to select all fields in the table
724 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
725 2379 aaronmk
        * container can be any iterable type
726 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
727
        * compare_right_side: sql_gen.ValueCond|literal value
728 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
729
        use all columns
730 2786 aaronmk
    @return query
731 1981 aaronmk
    '''
732 2315 aaronmk
    # Parse tables param
733 2964 aaronmk
    tables = lists.mk_seq(tables)
734 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
735 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
736 2121 aaronmk
737 2315 aaronmk
    # Parse other params
738 2376 aaronmk
    if conds == None: conds = []
739 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
740 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
741 3129 aaronmk
    assert limit == None or isinstance(limit, (int, long))
742
    assert start == None or isinstance(start, (int, long))
743 5527 aaronmk
    if limit == 0: order_by = None
744 2315 aaronmk
    if order_by is order_by_pkey:
745 5641 aaronmk
        if lists.is_seq(distinct_on) and distinct_on: order_by = distinct_on[0]
746
        elif table0 != None: order_by = table_order_by(db, table0, recover=True)
747
        else: order_by = None
748 865 aaronmk
749 2315 aaronmk
    query = 'SELECT'
750 2056 aaronmk
751 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
752 2056 aaronmk
753 2200 aaronmk
    # DISTINCT ON columns
754 2233 aaronmk
    if distinct_on != []:
755 2467 aaronmk
        query += '\nDISTINCT'
756 2254 aaronmk
        if distinct_on is not distinct_on_all:
757 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
758
759
    # Columns
760 3185 aaronmk
    if query.find('\n') >= 0: whitespace = '\n'
761
    else: whitespace = ' '
762
    if fields == None: query += whitespace+'*'
763 2765 aaronmk
    else:
764
        assert fields != []
765 3185 aaronmk
        if len(fields) > 1: whitespace = '\n'
766
        query += whitespace+('\n, '.join(map(parse_col, fields)))
767 2200 aaronmk
768
    # Main table
769 3185 aaronmk
    if query.find('\n') >= 0 or len(tables) > 0: whitespace = '\n'
770
    else: whitespace = ' '
771 3420 aaronmk
    if table0 != None: query += whitespace+'FROM '+table0.to_str(db)
772 865 aaronmk
773 2122 aaronmk
    # Add joins
774 2271 aaronmk
    left_table = table0
775 2263 aaronmk
    for join_ in tables:
776
        table = join_.table
777 2238 aaronmk
778 2343 aaronmk
        # Parse special values
779
        if join_.type_ is sql_gen.filter_out: # filter no match
780 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
781 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
782 2343 aaronmk
783 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
784 2122 aaronmk
785
        left_table = table
786
787 865 aaronmk
    missing = True
788 2376 aaronmk
    if conds != []:
789 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
790
        else: whitespace = '\n'
791 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
792
            .to_str(db) for l, r in conds], 'WHERE')
793 2227 aaronmk
    if order_by != None:
794 5642 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by).to_str(db)
795 3297 aaronmk
    if limit != None: query += '\nLIMIT '+str(limit)
796 865 aaronmk
    if start != None:
797 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
798 865 aaronmk
799 3494 aaronmk
    if explain: query = with_explain_comment(db, query)
800 3266 aaronmk
801 2786 aaronmk
    return query
802 11 aaronmk
803 2054 aaronmk
def select(db, *args, **kw_args):
804
    '''For params, see mk_select() and run_query()'''
805
    recover = kw_args.pop('recover', None)
806
    cacheable = kw_args.pop('cacheable', True)
807 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
808 2054 aaronmk
809 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
810
        log_level=log_level)
811 2054 aaronmk
812 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
813 3181 aaronmk
    embeddable=False, ignore=False, src=None):
814 1960 aaronmk
    '''
815
    @param returning str|None An inserted column (such as pkey) to return
816 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
817 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
818
        query will be fully cached, not just if it raises an exception.
819 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
820 3181 aaronmk
    @param src Will be included in the name of any created function, to help
821
        identify the data source in pg_stat_activity.
822 1960 aaronmk
    '''
823 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
824 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
825 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
826 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
827 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
828 2063 aaronmk
829 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
830 2063 aaronmk
831 3009 aaronmk
    def mk_insert(select_query):
832
        query = first_line
833 3014 aaronmk
        if cols != None:
834
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
835 3009 aaronmk
        query += '\n'+select_query
836
837
        if returning != None:
838
            returning_name_col = sql_gen.to_name_only_col(returning)
839
            query += '\nRETURNING '+returning_name_col.to_str(db)
840
841
        return query
842 2063 aaronmk
843 3489 aaronmk
    return_type = sql_gen.CustomCode('unknown')
844
    if returning != None: return_type = sql_gen.ColType(returning)
845 3017 aaronmk
846 3009 aaronmk
    if ignore:
847 3017 aaronmk
        # Always return something to set the correct rowcount
848
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
849
850 3009 aaronmk
        embeddable = True # must use function
851 3010 aaronmk
852 3450 aaronmk
        if cols == None: row = [sql_gen.Col(sql_gen.all_cols, 'row')]
853
        else: row = [sql_gen.Col(c.name, 'row') for c in cols]
854 3092 aaronmk
855 3484 aaronmk
        query = sql_gen.RowExcIgnore(sql_gen.RowType(table), select_query,
856 3497 aaronmk
            sql_gen.ReturnQuery(mk_insert(sql_gen.Values(row).to_str(db))),
857
            cols)
858 3009 aaronmk
    else: query = mk_insert(select_query)
859
860 2070 aaronmk
    if embeddable:
861
        # Create function
862 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
863 3181 aaronmk
        if src != None: function_name = src+': '+function_name
864 2189 aaronmk
        while True:
865
            try:
866 3451 aaronmk
                func = db.TempFunction(function_name)
867 3489 aaronmk
                def_ = sql_gen.FunctionDef(func, sql_gen.SetOf(return_type),
868
                    query)
869 2194 aaronmk
870 3443 aaronmk
                run_query(db, def_.to_str(db), recover=True, cacheable=True,
871 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
872 2189 aaronmk
                break # this version was successful
873 2945 aaronmk
            except DuplicateException, e:
874 2189 aaronmk
                function_name = next_version(function_name)
875
                # try again with next version of name
876 2070 aaronmk
877 2337 aaronmk
        # Return query that uses function
878 3009 aaronmk
        cols = None
879
        if returning != None: cols = [returning]
880 3451 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(func), cols)
881
            # AS clause requires function alias
882 3298 aaronmk
        return mk_select(db, func_table, order_by=None)
883 2070 aaronmk
884 2787 aaronmk
    return query
885 2066 aaronmk
886 3074 aaronmk
def insert_select(db, table, *args, **kw_args):
887 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
888 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
889
        values in
890 2072 aaronmk
    '''
891 3141 aaronmk
    returning = kw_args.get('returning', None)
892
    ignore = kw_args.get('ignore', False)
893
894 2386 aaronmk
    into = kw_args.pop('into', None)
895
    if into != None: kw_args['embeddable'] = True
896 2066 aaronmk
    recover = kw_args.pop('recover', None)
897 3141 aaronmk
    if ignore: recover = True
898 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
899 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
900 2066 aaronmk
901 3141 aaronmk
    rowcount_only = ignore and returning == None # keep NULL rows on server
902
    if rowcount_only: into = sql_gen.Table('rowcount')
903
904 3074 aaronmk
    cur = run_query_into(db, mk_insert_select(db, table, *args, **kw_args),
905
        into, recover=recover, cacheable=cacheable, log_level=log_level)
906 3141 aaronmk
    if rowcount_only: empty_temp(db, into)
907 3074 aaronmk
    autoanalyze(db, table)
908
    return cur
909 2063 aaronmk
910 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
911 2066 aaronmk
912 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
913 2085 aaronmk
    '''For params, see insert_select()'''
914 5050 aaronmk
    ignore = kw_args.pop('ignore', False)
915 5094 aaronmk
    if ignore: kw_args.setdefault('recover', True)
916 5050 aaronmk
917 1960 aaronmk
    if lists.is_seq(row): cols = None
918
    else:
919
        cols = row.keys()
920
        row = row.values()
921 2738 aaronmk
    row = list(row) # ensure that "== []" works
922 1960 aaronmk
923 2738 aaronmk
    if row == []: query = None
924
    else: query = sql_gen.Values(row).to_str(db)
925 1961 aaronmk
926 5050 aaronmk
    try: return insert_select(db, table, cols, query, *args, **kw_args)
927 5057 aaronmk
    except (DuplicateKeyException, NullValueException):
928 5050 aaronmk
        if not ignore: raise
929 5163 aaronmk
        return None
930 11 aaronmk
931 3152 aaronmk
def mk_update(db, table, changes=None, cond=None, in_place=False,
932 3153 aaronmk
    cacheable_=True):
933 2402 aaronmk
    '''
934
    @param changes [(col, new_value),...]
935
        * container can be any iterable type
936
        * col: sql_gen.Code|str (for col name)
937
        * new_value: sql_gen.Code|literal value
938
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
939 3056 aaronmk
    @param in_place If set, locks the table and updates rows in place.
940
        This avoids creating dead rows in PostgreSQL.
941
        * cond must be None
942 3153 aaronmk
    @param cacheable_ Whether column structure information used to generate the
943 3152 aaronmk
        query can be cached
944 2402 aaronmk
    @return str query
945
    '''
946 3057 aaronmk
    table = sql_gen.as_Table(table)
947
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
948
        for c, v in changes]
949
950 3056 aaronmk
    if in_place:
951
        assert cond == None
952 3058 aaronmk
953 5398 aaronmk
        def col_type(col):
954
            return sql_gen.canon_type(db.col_info(
955
                sql_gen.with_default_table(c, table), cacheable_).type)
956
        changes = [(c, v, col_type(c)) for c, v in changes]
957 3065 aaronmk
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
958 5396 aaronmk
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '+t+'\nUSING '
959
            +v.to_str(db) for c, v, t in changes))
960 3058 aaronmk
    else:
961
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
962
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
963
            for c, v in changes))
964
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
965 3056 aaronmk
966 3266 aaronmk
    query = with_explain_comment(db, query)
967
968 2402 aaronmk
    return query
969
970 3074 aaronmk
def update(db, table, *args, **kw_args):
971 2402 aaronmk
    '''For params, see mk_update() and run_query()'''
972
    recover = kw_args.pop('recover', None)
973 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
974 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
975 2402 aaronmk
976 3074 aaronmk
    cur = run_query(db, mk_update(db, table, *args, **kw_args), recover,
977
        cacheable, log_level=log_level)
978
    autoanalyze(db, table)
979
    return cur
980 2402 aaronmk
981 3286 aaronmk
def mk_delete(db, table, cond=None):
982
    '''
983
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
984
    @return str query
985
    '''
986
    query = 'DELETE FROM '+table.to_str(db)
987
    if cond != None: query += '\nWHERE '+cond.to_str(db)
988
989
    query = with_explain_comment(db, query)
990
991
    return query
992
993
def delete(db, table, *args, **kw_args):
994
    '''For params, see mk_delete() and run_query()'''
995
    recover = kw_args.pop('recover', None)
996 3295 aaronmk
    cacheable = kw_args.pop('cacheable', True)
997 3286 aaronmk
    log_level = kw_args.pop('log_level', 2)
998
999
    cur = run_query(db, mk_delete(db, table, *args, **kw_args), recover,
1000
        cacheable, log_level=log_level)
1001
    autoanalyze(db, table)
1002
    return cur
1003
1004 135 aaronmk
def last_insert_id(db):
1005 1849 aaronmk
    module = util.root_module(db.db)
1006 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
1007
    elif module == 'MySQLdb': return db.insert_id()
1008
    else: return None
1009 13 aaronmk
1010 3490 aaronmk
def define_func(db, def_):
1011
    func = def_.function
1012
    while True:
1013
        try:
1014
            run_query(db, def_.to_str(db), recover=True, cacheable=True,
1015
                log_ignore_excs=(DuplicateException,))
1016
            break # successful
1017
        except DuplicateException:
1018 3495 aaronmk
            func.name = next_version(func.name)
1019 3490 aaronmk
            # try again with next version of name
1020
1021 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
1022 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
1023 2415 aaronmk
    to names that will be distinct among the columns' tables.
1024 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
1025 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
1026
    @param into The table for the new columns.
1027 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
1028
        columns will be included in the mapping even if they are not in cols.
1029
        The tables of the provided Col objects will be changed to into, so make
1030
        copies of them if you want to keep the original tables.
1031
    @param as_items Whether to return a list of dict items instead of a dict
1032 2383 aaronmk
    @return dict(orig_col=new_col, ...)
1033
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
1034 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
1035
        * All mappings use the into table so its name can easily be
1036 2383 aaronmk
          changed for all columns at once
1037
    '''
1038 2415 aaronmk
    cols = lists.uniqify(cols)
1039
1040 2394 aaronmk
    items = []
1041 2389 aaronmk
    for col in preserve:
1042 2390 aaronmk
        orig_col = copy.copy(col)
1043 2392 aaronmk
        col.table = into
1044 2394 aaronmk
        items.append((orig_col, col))
1045
    preserve = set(preserve)
1046
    for col in cols:
1047 2716 aaronmk
        if col not in preserve:
1048 3750 aaronmk
            items.append((col, sql_gen.Col(strings.ustr(col), into, col.srcs)))
1049 2394 aaronmk
1050
    if not as_items: items = dict(items)
1051
    return items
1052 2383 aaronmk
1053 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
1054 2391 aaronmk
    '''For params, see mk_flatten_mapping()
1055
    @return See return value of mk_flatten_mapping()
1056
    '''
1057 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
1058
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
1059 5523 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
1060
        into=into, add_pkey_=True)
1061 3708 aaronmk
        # don't cache because the temp table will usually be truncated after use
1062 2394 aaronmk
    return dict(items)
1063 2391 aaronmk
1064 3079 aaronmk
##### Database structure introspection
1065 2414 aaronmk
1066 3079 aaronmk
#### Tables
1067
1068 4555 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False,
1069
    cacheable=True):
1070 3079 aaronmk
    if exact: compare = '='
1071
    else: compare = 'LIKE'
1072
1073
    module = util.root_module(db.db)
1074
    if module == 'psycopg2':
1075
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1076
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1077
        return values(select(db, 'pg_tables', ['tablename'], conds,
1078 4555 aaronmk
            order_by='tablename', cacheable=cacheable, log_level=4))
1079 3079 aaronmk
    elif module == 'MySQLdb':
1080
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1081
            , cacheable=True, log_level=4))
1082
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1083
1084 4556 aaronmk
def table_exists(db, table, cacheable=True):
1085 3079 aaronmk
    table = sql_gen.as_Table(table)
1086 4556 aaronmk
    return list(tables(db, table.schema, table.name, True, cacheable)) != []
1087 3079 aaronmk
1088 2426 aaronmk
def table_row_count(db, table, recover=None):
1089 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
1090 3298 aaronmk
        order_by=None), recover=recover, log_level=3))
1091 2426 aaronmk
1092 5337 aaronmk
def table_col_names(db, table, recover=None):
1093 5528 aaronmk
    return list(col_names(select(db, table, limit=0, recover=recover,
1094
        log_level=4)))
1095 2414 aaronmk
1096 5383 aaronmk
def table_cols(db, table, *args, **kw_args):
1097
    return [sql_gen.as_Col(strings.ustr(c), table)
1098
        for c in table_col_names(db, table, *args, **kw_args)]
1099
1100 5521 aaronmk
def table_pkey_index(db, table, recover=None):
1101
    table_str = sql_gen.Literal(table.to_str(db))
1102
    try:
1103
        return sql_gen.Table(value(run_query(db, '''\
1104
SELECT relname
1105
FROM pg_index
1106
JOIN pg_class index ON index.oid = indexrelid
1107
WHERE
1108
indrelid = '''+table_str.to_str(db)+'''::regclass
1109
AND indisprimary
1110
'''
1111
            , recover, cacheable=True, log_level=4)), table.schema)
1112
    except StopIteration: raise DoesNotExistException('primary key', '')
1113
1114 5389 aaronmk
def table_pkey_col(db, table, recover=None):
1115 5061 aaronmk
    table = sql_gen.as_Table(table)
1116
1117
    join_cols = ['table_schema', 'table_name', 'constraint_schema',
1118
        'constraint_name']
1119
    tables = [sql_gen.Table('key_column_usage', 'information_schema'),
1120
        sql_gen.Join(sql_gen.Table('table_constraints', 'information_schema'),
1121
            dict(((c, sql_gen.join_same_not_null) for c in join_cols)))]
1122
    cols = [sql_gen.Col('column_name')]
1123
1124
    conds = [('constraint_type', 'PRIMARY KEY'), ('table_name', table.name)]
1125
    schema = table.schema
1126
    if schema != None: conds.append(('table_schema', schema))
1127
    order_by = 'position_in_unique_constraint'
1128
1129 5393 aaronmk
    try: return sql_gen.Col(value(select(db, tables, cols, conds,
1130
        order_by=order_by, limit=1, log_level=4)), table)
1131
    except StopIteration: raise DoesNotExistException('primary key', '')
1132 5389 aaronmk
1133
def pkey_name(db, table, recover=None):
1134
    '''If no pkey, returns the first column in the table.'''
1135 5392 aaronmk
    return pkey_col(db, table, recover).name
1136 832 aaronmk
1137 5390 aaronmk
def pkey_col(db, table, recover=None):
1138 5391 aaronmk
    '''If no pkey, returns the first column in the table.'''
1139 5392 aaronmk
    try: return table_pkey_col(db, table, recover)
1140 5393 aaronmk
    except DoesNotExistException: return table_cols(db, table, recover)[0]
1141 5128 aaronmk
1142 2559 aaronmk
not_null_col = 'not_null_col'
1143 2340 aaronmk
1144
def table_not_null_col(db, table, recover=None):
1145
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
1146 5337 aaronmk
    if not_null_col in table_col_names(db, table, recover): return not_null_col
1147 5388 aaronmk
    else: return pkey_name(db, table, recover)
1148 2340 aaronmk
1149 3348 aaronmk
def constraint_cond(db, constraint):
1150
    module = util.root_module(db.db)
1151
    if module == 'psycopg2':
1152
        table_str = sql_gen.Literal(constraint.table.to_str(db))
1153
        name_str = sql_gen.Literal(constraint.name)
1154
        return value(run_query(db, '''\
1155
SELECT consrc
1156
FROM pg_constraint
1157
WHERE
1158
conrelid = '''+table_str.to_str(db)+'''::regclass
1159
AND conname = '''+name_str.to_str(db)+'''
1160
'''
1161
            , cacheable=True, log_level=4))
1162 5443 aaronmk
    else: raise NotImplementedError("Can't get constraint condition for "
1163
        +module+' database')
1164 3348 aaronmk
1165 5520 aaronmk
def index_exprs(db, index):
1166 3322 aaronmk
    index = sql_gen.as_Table(index)
1167 1909 aaronmk
    module = util.root_module(db.db)
1168
    if module == 'psycopg2':
1169 3322 aaronmk
        qual_index = sql_gen.Literal(index.to_str(db))
1170 5520 aaronmk
        return list(values(run_query(db, '''\
1171 3322 aaronmk
SELECT pg_get_indexdef(indexrelid, generate_series(1, indnatts), true)
1172
FROM pg_index
1173
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1174 2782 aaronmk
'''
1175
            , cacheable=True, log_level=4)))
1176 5520 aaronmk
    else: raise NotImplementedError()
1177 853 aaronmk
1178 5520 aaronmk
def index_cols(db, index):
1179
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
1180
    automatically created. When you don't know whether something is a UNIQUE
1181
    constraint or a UNIQUE index, use this function.'''
1182
    return map(sql_gen.parse_expr_col, index_exprs(db, index))
1183
1184 5445 aaronmk
def index_cond(db, index):
1185
    index = sql_gen.as_Table(index)
1186
    module = util.root_module(db.db)
1187
    if module == 'psycopg2':
1188
        qual_index = sql_gen.Literal(index.to_str(db))
1189
        return value(run_query(db, '''\
1190
SELECT pg_get_expr(indpred, indrelid, true)
1191
FROM pg_index
1192
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1193
'''
1194
            , cacheable=True, log_level=4))
1195
    else: raise NotImplementedError()
1196
1197 5521 aaronmk
def index_order_by(db, index):
1198
    return sql_gen.CustomCode(', '.join(index_exprs(db, index)))
1199
1200
def table_cluster_on(db, table, recover=None):
1201
    '''
1202
    @return The table's cluster index, or its pkey if none is set
1203
    '''
1204
    table_str = sql_gen.Literal(table.to_str(db))
1205
    try:
1206
        return sql_gen.Table(value(run_query(db, '''\
1207
SELECT relname
1208
FROM pg_index
1209
JOIN pg_class index ON index.oid = indexrelid
1210
WHERE
1211
indrelid = '''+table_str.to_str(db)+'''::regclass
1212
AND indisclustered
1213
'''
1214
            , recover, cacheable=True, log_level=4)), table.schema)
1215
    except StopIteration: return table_pkey_index(db, table, recover)
1216
1217
def table_order_by(db, table, recover=None):
1218 5525 aaronmk
    if table.order_by == None:
1219
        try: table.order_by = index_order_by(db, table_cluster_on(db, table,
1220
            recover))
1221
        except DoesNotExistException: pass
1222
    return table.order_by
1223 5521 aaronmk
1224 3079 aaronmk
#### Functions
1225
1226
def function_exists(db, function):
1227 3423 aaronmk
    qual_function = sql_gen.Literal(function.to_str(db))
1228
    try:
1229 3425 aaronmk
        select(db, fields=[sql_gen.Cast('regproc', qual_function)],
1230
            recover=True, cacheable=True, log_level=4)
1231 3423 aaronmk
    except DoesNotExistException: return False
1232 4146 aaronmk
    except DuplicateException: return True # overloaded function
1233 3423 aaronmk
    else: return True
1234 3079 aaronmk
1235 5713 aaronmk
def function_param0_type(db, function):
1236
    qual_function = sql_gen.Literal(function.to_str(db))
1237
    return value(run_query(db, '''\
1238
SELECT proargtypes[0]::regtype
1239
FROM pg_proc
1240
WHERE oid = '''+qual_function.to_str(db)+'''::regproc
1241
'''
1242
        , cacheable=True, log_level=4))
1243
1244 3079 aaronmk
##### Structural changes
1245
1246
#### Columns
1247
1248 5020 aaronmk
def add_col(db, table, col, comment=None, if_not_exists=False, **kw_args):
1249 3079 aaronmk
    '''
1250
    @param col TypedCol Name may be versioned, so be sure to propagate any
1251
        renaming back to any source column for the TypedCol.
1252
    @param comment None|str SQL comment used to distinguish columns of the same
1253
        name from each other when they contain different data, to allow the
1254
        ADD COLUMN query to be cached. If not set, query will not be cached.
1255
    '''
1256
    assert isinstance(col, sql_gen.TypedCol)
1257
1258
    while True:
1259
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1260
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1261
1262
        try:
1263
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1264
            break
1265
        except DuplicateException:
1266 5020 aaronmk
            if if_not_exists: raise
1267 3079 aaronmk
            col.name = next_version(col.name)
1268
            # try again with next version of name
1269
1270
def add_not_null(db, col):
1271
    table = col.table
1272
    col = sql_gen.to_name_only_col(col)
1273
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1274
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1275
1276 4443 aaronmk
def drop_not_null(db, col):
1277
    table = col.table
1278
    col = sql_gen.to_name_only_col(col)
1279
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1280
        +col.to_str(db)+' DROP NOT NULL', cacheable=True, log_level=3)
1281
1282 2096 aaronmk
row_num_col = '_row_num'
1283
1284 4997 aaronmk
row_num_col_def = sql_gen.TypedCol('', 'serial', nullable=False,
1285 3079 aaronmk
    constraints='PRIMARY KEY')
1286
1287 4997 aaronmk
def add_row_num(db, table, name=row_num_col):
1288
    '''Adds a row number column to a table. Its definition is in
1289
    row_num_col_def. It will be the primary key.'''
1290
    col_def = copy.copy(row_num_col_def)
1291
    col_def.name = name
1292 5021 aaronmk
    add_col(db, table, col_def, comment='', if_not_exists=True, log_level=3)
1293 3079 aaronmk
1294
#### Indexes
1295
1296
def add_pkey(db, table, cols=None, recover=None):
1297
    '''Adds a primary key.
1298
    @param cols [sql_gen.Col,...] The columns in the primary key.
1299
        Defaults to the first column in the table.
1300
    @pre The table must not already have a primary key.
1301
    '''
1302
    table = sql_gen.as_Table(table)
1303 5388 aaronmk
    if cols == None: cols = [pkey_name(db, table, recover)]
1304 3079 aaronmk
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1305
1306
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1307
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1308
        log_ignore_excs=(DuplicateException,))
1309
1310 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1311 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1312 3356 aaronmk
    Currently, only function calls and literal values are supported expressions.
1313 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1314 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1315 2538 aaronmk
    '''
1316 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1317 2538 aaronmk
1318 2688 aaronmk
    # Parse exprs
1319
    old_exprs = exprs[:]
1320
    exprs = []
1321
    cols = []
1322
    for i, expr in enumerate(old_exprs):
1323 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1324 2688 aaronmk
1325 2823 aaronmk
        # Handle nullable columns
1326 2998 aaronmk
        if ensure_not_null_:
1327 3164 aaronmk
            try: expr = sql_gen.ensure_not_null(db, expr)
1328 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1329 2823 aaronmk
1330 2688 aaronmk
        # Extract col
1331 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1332 3356 aaronmk
        col = expr
1333
        if isinstance(expr, sql_gen.FunctionCall): col = expr.args[0]
1334
        expr = sql_gen.cast_literal(expr)
1335
        if not isinstance(expr, (sql_gen.Expr, sql_gen.Col)):
1336 2688 aaronmk
            expr = sql_gen.Expr(expr)
1337 3356 aaronmk
1338 2688 aaronmk
1339
        # Extract table
1340
        if table == None:
1341
            assert sql_gen.is_table_col(col)
1342
            table = col.table
1343
1344 3356 aaronmk
        if isinstance(col, sql_gen.Col): col.table = None
1345 2688 aaronmk
1346
        exprs.append(expr)
1347
        cols.append(col)
1348 2408 aaronmk
1349 2688 aaronmk
    table = sql_gen.as_Table(table)
1350
1351 3005 aaronmk
    # Add index
1352 3148 aaronmk
    str_ = 'CREATE'
1353
    if unique: str_ += ' UNIQUE'
1354
    str_ += ' INDEX ON '+table.to_str(db)+' ('+(
1355
        ', '.join((v.to_str(db) for v in exprs)))+')'
1356
    run_query(db, str_, recover=True, cacheable=True, log_level=3)
1357 2408 aaronmk
1358 5765 aaronmk
def add_pkey_or_index(db, table, cols=None, recover=None, warn=False):
1359
    try: add_pkey(db, table, cols, recover)
1360
    except DuplicateKeyException, e:
1361
        if warn: warnings.warn(UserWarning(exc.str_(e)))
1362
        add_index(db, pkey_col(db, table), table)
1363
1364 3083 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1365
1366
def add_indexes(db, table, has_pkey=True):
1367
    '''Adds an index on all columns in a table.
1368
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1369
        index should be added on the first column.
1370
        * If already_indexed, the pkey is assumed to have already been added
1371
    '''
1372 5337 aaronmk
    cols = table_col_names(db, table)
1373 3083 aaronmk
    if has_pkey:
1374
        if has_pkey is not already_indexed: add_pkey(db, table)
1375
        cols = cols[1:]
1376
    for col in cols: add_index(db, col, table)
1377
1378 3079 aaronmk
#### Tables
1379 2772 aaronmk
1380 3079 aaronmk
### Maintenance
1381 2772 aaronmk
1382 3079 aaronmk
def analyze(db, table):
1383
    table = sql_gen.as_Table(table)
1384
    run_query(db, 'ANALYZE '+table.to_str(db), log_level=3)
1385 2934 aaronmk
1386 3079 aaronmk
def autoanalyze(db, table):
1387
    if db.autoanalyze: analyze(db, table)
1388 2935 aaronmk
1389 3079 aaronmk
def vacuum(db, table):
1390
    table = sql_gen.as_Table(table)
1391
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1392
        log_level=3))
1393 2086 aaronmk
1394 3079 aaronmk
### Lifecycle
1395
1396 3247 aaronmk
def drop(db, type_, name):
1397
    name = sql_gen.as_Name(name)
1398
    run_query(db, 'DROP '+type_+' IF EXISTS '+name.to_str(db)+' CASCADE')
1399 2889 aaronmk
1400 3247 aaronmk
def drop_table(db, table): drop(db, 'TABLE', table)
1401
1402 3082 aaronmk
def create_table(db, table, cols=[], has_pkey=True, col_indexes=True,
1403
    like=None):
1404 2675 aaronmk
    '''Creates a table.
1405 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1406
    @param has_pkey If set, the first column becomes the primary key.
1407 2760 aaronmk
    @param col_indexes bool|[ref]
1408
        * If True, indexes will be added on all non-pkey columns.
1409
        * If a list reference, [0] will be set to a function to do this.
1410
          This can be used to delay index creation until the table is populated.
1411 2675 aaronmk
    '''
1412
    table = sql_gen.as_Table(table)
1413
1414 3082 aaronmk
    if like != None:
1415
        cols = [sql_gen.CustomCode('LIKE '+like.to_str(db)+' INCLUDING ALL')
1416
            ]+cols
1417 5525 aaronmk
        table.order_by = like.order_by
1418 2681 aaronmk
    if has_pkey:
1419
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1420 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1421 2681 aaronmk
1422 3085 aaronmk
    temp = table.is_temp and not db.debug_temp
1423
        # temp tables permanent in debug_temp mode
1424 2760 aaronmk
1425 3085 aaronmk
    # Create table
1426 3383 aaronmk
    def create():
1427 3085 aaronmk
        str_ = 'CREATE'
1428
        if temp: str_ += ' TEMP'
1429
        str_ += ' TABLE '+table.to_str(db)+' (\n'
1430
        str_ += '\n, '.join(c.to_str(db) for c in cols)
1431 3126 aaronmk
        str_ += '\n);'
1432 3085 aaronmk
1433 3383 aaronmk
        run_query(db, str_, recover=True, cacheable=True, log_level=2,
1434
            log_ignore_excs=(DuplicateException,))
1435
    if table.is_temp:
1436
        while True:
1437
            try:
1438
                create()
1439
                break
1440
            except DuplicateException:
1441
                table.name = next_version(table.name)
1442
                # try again with next version of name
1443
    else: create()
1444 3085 aaronmk
1445 2760 aaronmk
    # Add indexes
1446 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1447
    def add_indexes_(): add_indexes(db, table, has_pkey)
1448
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1449
    elif col_indexes: add_indexes_() # add now
1450 2675 aaronmk
1451 3084 aaronmk
def copy_table_struct(db, src, dest):
1452
    '''Creates a structure-only copy of a table. (Does not copy data.)'''
1453 3085 aaronmk
    create_table(db, dest, has_pkey=False, col_indexes=False, like=src)
1454 3084 aaronmk
1455 5529 aaronmk
def copy_table(db, src, dest):
1456
    '''Creates a copy of a table, including data'''
1457
    copy_table_struct(db, src, dest)
1458
    insert_select(db, dest, None, mk_select(db, src))
1459
1460 3079 aaronmk
### Data
1461 2684 aaronmk
1462 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1463
    '''For params, see run_query()'''
1464 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1465 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1466 2732 aaronmk
1467 2965 aaronmk
def empty_temp(db, tables):
1468
    tables = lists.mk_seq(tables)
1469 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1470 2965 aaronmk
1471 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1472
    '''For kw_args, see tables()'''
1473
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1474 3094 aaronmk
1475
def distinct_table(db, table, distinct_on):
1476
    '''Creates a copy of a temp table which is distinct on the given columns.
1477 5902 aaronmk
    Adds an index on table's distinct_on columns, to facilitate merge joins.
1478 3097 aaronmk
    @param distinct_on If empty, creates a table with one row. This is useful if
1479
        your distinct_on columns are all literal values.
1480 3099 aaronmk
    @return The new table.
1481 3094 aaronmk
    '''
1482 3099 aaronmk
    new_table = sql_gen.suffixed_table(table, '_distinct')
1483 3411 aaronmk
    distinct_on = filter(sql_gen.is_table_col, distinct_on)
1484 3094 aaronmk
1485 3099 aaronmk
    copy_table_struct(db, table, new_table)
1486 3097 aaronmk
1487
    limit = None
1488
    if distinct_on == []: limit = 1 # one sample row
1489 5903 aaronmk
    else: add_index(db, distinct_on, table) # for join optimization
1490 3097 aaronmk
1491 5903 aaronmk
    insert_select(db, new_table, None, mk_select(db, table,
1492
        distinct_on=distinct_on, order_by=None, limit=limit))
1493 3099 aaronmk
    analyze(db, new_table)
1494 3094 aaronmk
1495 3099 aaronmk
    return new_table